G-BLASTN

G-BLASTN accelerates nucleotide sequence alignment by leveraging Graphics Processing Units (GPUs) to implement NCBI-BLAST megablast and blastn searches for genome-scale sequence analysis.

Key Features:

• GPU Acceleration: Uses GPUs to parallelize nucleotide alignment computations for megablast and blastn modes.
• Compatibility and Accuracy: Produces results identical to NCBI-BLAST.
• Performance Gains: Under 'megablast' mode achieves a 14.80× speedup versus sequential NCBI-BLAST; versus a multithreaded NCBI-BLAST on four CPU cores provides 7.15× speedup in 'megablast' mode and 4.32× in 'blastn' mode; compared to single-core execution provides a 1.56× speedup in 'blastn' mode.
• Pipeline Mode: Includes a pipeline mode that improves batch processing performance by up to 44% for multiple-query workloads.
• Optimization Focus: Currently optimized for databases containing long sequences and not yet optimized for short database sequences.

Scientific Applications:

• Genome assembly: Speeds nucleotide alignment steps in genome assembly workflows.
• Genome annotation: Accelerates homology searches used in genome annotation.
• Comparative genomics: Enables rapid alignments for comparative genomics analyses.
• Variant identification: Supports identification of genetic variations through faster nucleotide alignments.

Methodology:

Implements a GPU-accelerated reimplementation of NCBI-BLAST algorithms supporting megablast and blastn modes and a pipeline mode for batch query processing.